Individual differences of neurophysiological brain activity are heritable

Introduction

Brain-fingerprint is a novel technique used to explore the inter-individual differrences in the brain activity. Previous research demonstrates that we are capable of accurately differentiating individuals in a cohort based solely on the resting-stage spectral brain activy using magnetoencephalography (MEG). How these fingerprints are influenced by genetics remains to be demonstrated. Here, we explored the heritablility of brain-fingerprints and their alignment with cortical patterns of gene expression.

Description of the project

Here are the different steps we did :

• We apply the brain-fingerprint method to diffenrentiate both individuals and twin pairs : PSD_accuracy.py
• To compare with chance-level, we followed the same pipeline, but with the empty room recording : PSD_accuracy_empty_room.py
• We futhermore explored the same method, but restricting ourselves to shorter segments of recording : PSD_accuracy_30_sec.py
• We observed the correlations between individuals, and the distributions of correlation within twin pairs : PSD_correlations.py
• To evaluate how useful are the different features from the spectral brain for fingerprinting, we computed the intraclass correlation (ICC) : PSD_ICC_Fingerprint.py
• We computed then the heritability of the same features, using the Falconer's formula : PSD_Heritability.py
• We also computed the the heritability of the anatomical features, to see the influence of anatomy on the spectrel-brain heritability : PSD_Heritability_Anatomy.py

All of those steps are merged together in main.py

Statistical analyses and plots were performed : plots_Schaefer.R

We additionally ran a PLS analysis to relate salient features for participant differentiation (ICC) to gene expression from the AHBA (see Hansen et al 2021. Code for the PLS analysis can be found in the PLS_Code folder.

Results

In short we found that:

• Monozygotic, but not dizygotic twins can be differentiated from their sibling's brain-fingerprint
• brain-fingerprints, principally in the alpha and beta bands, are heritable
• salient electrophysiological features for participant differentiation coavry with a gradient of gene expression
• this gradient of gene expression is enriched for ion transport genes, expressed in neurons, and becomes more pronounced across neurodevelopment

Manuscript and Citation

This work on bioRxiv. Please cite da Silva Castanheira et al., 2024. If you have any questions please contact the authors of the paper.

How to use this repository

In addition to the provided data on this github repository, you will require :

• to request the access to the resticted data for the dataset for the MEG subjects, and add the non-resticted data to the folder "Data" as "All_Data.csv", as well as the restricted dataset containing the information for the twin pairs, to name "All_Data_RESTRICTED.csv".
• to download the "Schaefer_30_second" folder if you want to run the analysis using the short segments and to upload it in "Data
• to download the "Schaefer_artifacts_correction" folder if you want to run the analysis using the PSDs after regressing out the artifacts

Note scripts for our PLS analyses can be found in the PLS codes folder. R scripts provide useful plotting functions for data visualization. All the results to conduct the brain-fingerprinting ananlyses can be computed using main.py :

Example :

python3 main.py --acc True -- only_gt True 

Here are the different parameters :

• --acc : either if we compute the accuracy or not
• --acc_30_sec : either if we compute the accuracy or not for the 30-sec PSDs
• --acc_empty_room : either if we compute the accuracy of the empty room or not
• --corr : either if we compute the correlations or not
• --heritability : either if we compute the heritability or not
• --heritability_anatomy : either if we compute the hertiability of anatomy or not
• --fingerprint : either if we compute the ICC for fingerprinting or not
• --data : path of the folder containing the data
• --data_30_sec : path of the folder containing the 30-seconds PSDs
• --results : path for the results
• --only_gt : choice of if we use only twin pairs based on genetic test ("True"), or if we also use the self reported twins ("False")
• --correlation_type : correlation used for heritability (pearson or icc)
• --n_resample : number of bootstraps
• --icc_without_twins : either ICC for fingerprinting is computed with or without twins